snowpark-connect 0.24.0__py3-none-any.whl → 0.26.0__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Potentially problematic release.


This version of snowpark-connect might be problematic. Click here for more details.

Files changed (484) hide show
  1. snowflake/snowpark_connect/column_name_handler.py +116 -4
  2. snowflake/snowpark_connect/config.py +23 -0
  3. snowflake/snowpark_connect/constants.py +0 -29
  4. snowflake/snowpark_connect/dataframe_container.py +22 -0
  5. snowflake/snowpark_connect/execute_plan/map_execution_command.py +56 -1
  6. snowflake/snowpark_connect/expression/literal.py +13 -2
  7. snowflake/snowpark_connect/expression/map_cast.py +5 -8
  8. snowflake/snowpark_connect/expression/map_sql_expression.py +23 -1
  9. snowflake/snowpark_connect/expression/map_udf.py +88 -29
  10. snowflake/snowpark_connect/expression/map_unresolved_attribute.py +199 -15
  11. snowflake/snowpark_connect/expression/map_unresolved_extract_value.py +44 -16
  12. snowflake/snowpark_connect/expression/map_unresolved_function.py +840 -367
  13. snowflake/snowpark_connect/expression/map_unresolved_star.py +3 -2
  14. snowflake/snowpark_connect/hidden_column.py +39 -0
  15. snowflake/snowpark_connect/includes/jars/hadoop-client-api-trimmed-3.3.4.jar +0 -0
  16. snowflake/snowpark_connect/includes/jars/json4s-native_2.12-3.7.0-M11.jar +0 -0
  17. snowflake/snowpark_connect/includes/jars/paranamer-2.8.3.jar +0 -0
  18. snowflake/snowpark_connect/includes/jars/sas-scala-udf_2.12-0.1.0.jar +0 -0
  19. snowflake/snowpark_connect/includes/jars/{hadoop-client-api-3.3.4.jar → spark-connect-client-jvm_2.12-3.5.6.jar} +0 -0
  20. snowflake/snowpark_connect/relation/map_column_ops.py +17 -4
  21. snowflake/snowpark_connect/relation/map_extension.py +52 -11
  22. snowflake/snowpark_connect/relation/map_join.py +258 -62
  23. snowflake/snowpark_connect/relation/map_map_partitions.py +9 -4
  24. snowflake/snowpark_connect/relation/map_relation.py +12 -1
  25. snowflake/snowpark_connect/relation/map_row_ops.py +8 -1
  26. snowflake/snowpark_connect/relation/map_sql.py +88 -11
  27. snowflake/snowpark_connect/relation/map_udtf.py +100 -46
  28. snowflake/snowpark_connect/relation/read/map_read.py +3 -3
  29. snowflake/snowpark_connect/relation/read/map_read_jdbc.py +1 -1
  30. snowflake/snowpark_connect/relation/read/map_read_json.py +8 -1
  31. snowflake/snowpark_connect/relation/read/map_read_table.py +1 -9
  32. snowflake/snowpark_connect/relation/read/reader_config.py +3 -1
  33. snowflake/snowpark_connect/relation/utils.py +44 -0
  34. snowflake/snowpark_connect/relation/write/map_write.py +175 -75
  35. snowflake/snowpark_connect/resources_initializer.py +47 -6
  36. snowflake/snowpark_connect/server.py +26 -4
  37. snowflake/snowpark_connect/type_mapping.py +29 -25
  38. snowflake/snowpark_connect/typed_column.py +14 -0
  39. snowflake/snowpark_connect/utils/artifacts.py +23 -0
  40. snowflake/snowpark_connect/utils/concurrent.py +4 -0
  41. snowflake/snowpark_connect/utils/context.py +6 -1
  42. snowflake/snowpark_connect/utils/external_udxf_cache.py +36 -0
  43. snowflake/snowpark_connect/utils/scala_udf_utils.py +596 -0
  44. snowflake/snowpark_connect/utils/session.py +4 -0
  45. snowflake/snowpark_connect/utils/telemetry.py +6 -17
  46. snowflake/snowpark_connect/utils/udf_helper.py +2 -0
  47. snowflake/snowpark_connect/utils/udf_utils.py +22 -1
  48. snowflake/snowpark_connect/utils/udtf_utils.py +1 -0
  49. snowflake/snowpark_connect/version.py +1 -1
  50. {snowpark_connect-0.24.0.dist-info → snowpark_connect-0.26.0.dist-info}/METADATA +1 -1
  51. snowpark_connect-0.26.0.dist-info/RECORD +481 -0
  52. snowflake/snowpark_connect/includes/jars/scala-compiler-2.12.18.jar +0 -0
  53. snowflake/snowpark_connect/includes/jars/spark-kubernetes_2.12-3.5.6.jar +0 -0
  54. snowflake/snowpark_connect/includes/jars/spark-mllib_2.12-3.5.6.jar +0 -0
  55. snowflake/snowpark_connect/includes/jars/spark-streaming_2.12-3.5.6.jar +0 -0
  56. snowflake/snowpark_connect/includes/python/pyspark/errors/tests/__init__.py +0 -16
  57. snowflake/snowpark_connect/includes/python/pyspark/errors/tests/test_errors.py +0 -60
  58. snowflake/snowpark_connect/includes/python/pyspark/ml/deepspeed/tests/test_deepspeed_distributor.py +0 -306
  59. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/__init__.py +0 -16
  60. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/connect/test_connect_classification.py +0 -53
  61. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/connect/test_connect_evaluation.py +0 -50
  62. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/connect/test_connect_feature.py +0 -43
  63. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/connect/test_connect_function.py +0 -114
  64. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/connect/test_connect_pipeline.py +0 -47
  65. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/connect/test_connect_summarizer.py +0 -43
  66. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/connect/test_connect_tuning.py +0 -46
  67. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/connect/test_legacy_mode_classification.py +0 -238
  68. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/connect/test_legacy_mode_evaluation.py +0 -194
  69. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/connect/test_legacy_mode_feature.py +0 -156
  70. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/connect/test_legacy_mode_pipeline.py +0 -184
  71. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/connect/test_legacy_mode_summarizer.py +0 -78
  72. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/connect/test_legacy_mode_tuning.py +0 -292
  73. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/connect/test_parity_torch_data_loader.py +0 -50
  74. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/connect/test_parity_torch_distributor.py +0 -152
  75. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/test_algorithms.py +0 -456
  76. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/test_base.py +0 -96
  77. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/test_dl_util.py +0 -186
  78. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/test_evaluation.py +0 -77
  79. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/test_feature.py +0 -401
  80. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/test_functions.py +0 -528
  81. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/test_image.py +0 -82
  82. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/test_linalg.py +0 -409
  83. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/test_model_cache.py +0 -55
  84. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/test_param.py +0 -441
  85. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/test_persistence.py +0 -546
  86. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/test_pipeline.py +0 -71
  87. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/test_stat.py +0 -52
  88. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/test_training_summary.py +0 -494
  89. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/test_util.py +0 -85
  90. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/test_wrapper.py +0 -138
  91. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/tuning/__init__.py +0 -16
  92. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/tuning/test_cv_io_basic.py +0 -151
  93. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/tuning/test_cv_io_nested.py +0 -97
  94. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/tuning/test_cv_io_pipeline.py +0 -143
  95. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/tuning/test_tuning.py +0 -551
  96. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/tuning/test_tvs_io_basic.py +0 -137
  97. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/tuning/test_tvs_io_nested.py +0 -96
  98. snowflake/snowpark_connect/includes/python/pyspark/ml/tests/tuning/test_tvs_io_pipeline.py +0 -142
  99. snowflake/snowpark_connect/includes/python/pyspark/ml/torch/tests/__init__.py +0 -16
  100. snowflake/snowpark_connect/includes/python/pyspark/ml/torch/tests/test_data_loader.py +0 -137
  101. snowflake/snowpark_connect/includes/python/pyspark/ml/torch/tests/test_distributor.py +0 -561
  102. snowflake/snowpark_connect/includes/python/pyspark/ml/torch/tests/test_log_communication.py +0 -172
  103. snowflake/snowpark_connect/includes/python/pyspark/mllib/tests/__init__.py +0 -16
  104. snowflake/snowpark_connect/includes/python/pyspark/mllib/tests/test_algorithms.py +0 -353
  105. snowflake/snowpark_connect/includes/python/pyspark/mllib/tests/test_feature.py +0 -192
  106. snowflake/snowpark_connect/includes/python/pyspark/mllib/tests/test_linalg.py +0 -680
  107. snowflake/snowpark_connect/includes/python/pyspark/mllib/tests/test_stat.py +0 -206
  108. snowflake/snowpark_connect/includes/python/pyspark/mllib/tests/test_streaming_algorithms.py +0 -471
  109. snowflake/snowpark_connect/includes/python/pyspark/mllib/tests/test_util.py +0 -108
  110. snowflake/snowpark_connect/includes/python/pyspark/pandas/spark/__init__.py +0 -16
  111. snowflake/snowpark_connect/includes/python/pyspark/pandas/spark/accessors.py +0 -1281
  112. snowflake/snowpark_connect/includes/python/pyspark/pandas/spark/functions.py +0 -203
  113. snowflake/snowpark_connect/includes/python/pyspark/pandas/spark/utils.py +0 -202
  114. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/__init__.py +0 -16
  115. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/computation/__init__.py +0 -16
  116. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/computation/test_any_all.py +0 -177
  117. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/computation/test_apply_func.py +0 -575
  118. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/computation/test_binary_ops.py +0 -235
  119. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/computation/test_combine.py +0 -653
  120. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/computation/test_compute.py +0 -463
  121. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/computation/test_corrwith.py +0 -86
  122. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/computation/test_cov.py +0 -151
  123. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/computation/test_cumulative.py +0 -139
  124. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/computation/test_describe.py +0 -458
  125. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/computation/test_eval.py +0 -86
  126. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/computation/test_melt.py +0 -202
  127. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/computation/test_missing_data.py +0 -520
  128. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/computation/test_pivot.py +0 -361
  129. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/__init__.py +0 -16
  130. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/computation/__init__.py +0 -16
  131. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/computation/test_parity_any_all.py +0 -40
  132. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/computation/test_parity_apply_func.py +0 -42
  133. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/computation/test_parity_binary_ops.py +0 -40
  134. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/computation/test_parity_combine.py +0 -37
  135. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/computation/test_parity_compute.py +0 -60
  136. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/computation/test_parity_corrwith.py +0 -40
  137. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/computation/test_parity_cov.py +0 -40
  138. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/computation/test_parity_cumulative.py +0 -90
  139. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/computation/test_parity_describe.py +0 -40
  140. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/computation/test_parity_eval.py +0 -40
  141. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/computation/test_parity_melt.py +0 -40
  142. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/computation/test_parity_missing_data.py +0 -42
  143. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/computation/test_parity_pivot.py +0 -37
  144. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/data_type_ops/__init__.py +0 -16
  145. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/data_type_ops/test_parity_base.py +0 -36
  146. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/data_type_ops/test_parity_binary_ops.py +0 -42
  147. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/data_type_ops/test_parity_boolean_ops.py +0 -47
  148. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/data_type_ops/test_parity_categorical_ops.py +0 -55
  149. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/data_type_ops/test_parity_complex_ops.py +0 -40
  150. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/data_type_ops/test_parity_date_ops.py +0 -47
  151. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/data_type_ops/test_parity_datetime_ops.py +0 -47
  152. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/data_type_ops/test_parity_null_ops.py +0 -42
  153. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/data_type_ops/test_parity_num_arithmetic.py +0 -43
  154. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/data_type_ops/test_parity_num_ops.py +0 -47
  155. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/data_type_ops/test_parity_num_reverse.py +0 -43
  156. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/data_type_ops/test_parity_string_ops.py +0 -47
  157. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/data_type_ops/test_parity_timedelta_ops.py +0 -47
  158. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/data_type_ops/test_parity_udt_ops.py +0 -40
  159. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/data_type_ops/testing_utils.py +0 -226
  160. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/diff_frames_ops/__init__.py +0 -16
  161. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/diff_frames_ops/test_parity_align.py +0 -39
  162. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/diff_frames_ops/test_parity_basic_slow.py +0 -55
  163. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/diff_frames_ops/test_parity_cov_corrwith.py +0 -39
  164. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/diff_frames_ops/test_parity_dot_frame.py +0 -39
  165. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/diff_frames_ops/test_parity_dot_series.py +0 -39
  166. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/diff_frames_ops/test_parity_index.py +0 -39
  167. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/diff_frames_ops/test_parity_series.py +0 -39
  168. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/diff_frames_ops/test_parity_setitem_frame.py +0 -43
  169. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/diff_frames_ops/test_parity_setitem_series.py +0 -43
  170. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/frame/__init__.py +0 -16
  171. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/frame/test_parity_attrs.py +0 -40
  172. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/frame/test_parity_constructor.py +0 -39
  173. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/frame/test_parity_conversion.py +0 -42
  174. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/frame/test_parity_reindexing.py +0 -42
  175. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/frame/test_parity_reshaping.py +0 -37
  176. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/frame/test_parity_spark.py +0 -40
  177. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/frame/test_parity_take.py +0 -42
  178. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/frame/test_parity_time_series.py +0 -48
  179. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/frame/test_parity_truncate.py +0 -40
  180. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/groupby/__init__.py +0 -16
  181. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/groupby/test_parity_aggregate.py +0 -40
  182. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/groupby/test_parity_apply_func.py +0 -41
  183. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/groupby/test_parity_cumulative.py +0 -67
  184. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/groupby/test_parity_describe.py +0 -40
  185. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/groupby/test_parity_groupby.py +0 -55
  186. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/groupby/test_parity_head_tail.py +0 -40
  187. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/groupby/test_parity_index.py +0 -38
  188. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/groupby/test_parity_missing_data.py +0 -55
  189. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/groupby/test_parity_split_apply.py +0 -39
  190. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/groupby/test_parity_stat.py +0 -38
  191. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/indexes/__init__.py +0 -16
  192. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/indexes/test_parity_align.py +0 -40
  193. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/indexes/test_parity_base.py +0 -50
  194. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/indexes/test_parity_category.py +0 -73
  195. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/indexes/test_parity_datetime.py +0 -39
  196. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/indexes/test_parity_indexing.py +0 -40
  197. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/indexes/test_parity_reindex.py +0 -40
  198. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/indexes/test_parity_rename.py +0 -40
  199. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/indexes/test_parity_reset_index.py +0 -48
  200. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/indexes/test_parity_timedelta.py +0 -39
  201. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/io/__init__.py +0 -16
  202. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/io/test_parity_io.py +0 -40
  203. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/plot/__init__.py +0 -16
  204. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/plot/test_parity_frame_plot.py +0 -45
  205. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/plot/test_parity_frame_plot_matplotlib.py +0 -45
  206. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/plot/test_parity_frame_plot_plotly.py +0 -49
  207. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/plot/test_parity_series_plot.py +0 -37
  208. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/plot/test_parity_series_plot_matplotlib.py +0 -53
  209. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/plot/test_parity_series_plot_plotly.py +0 -45
  210. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/series/__init__.py +0 -16
  211. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/series/test_parity_all_any.py +0 -38
  212. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/series/test_parity_arg_ops.py +0 -37
  213. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/series/test_parity_as_of.py +0 -37
  214. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/series/test_parity_as_type.py +0 -38
  215. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/series/test_parity_compute.py +0 -37
  216. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/series/test_parity_conversion.py +0 -40
  217. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/series/test_parity_cumulative.py +0 -40
  218. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/series/test_parity_index.py +0 -38
  219. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/series/test_parity_missing_data.py +0 -40
  220. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/series/test_parity_series.py +0 -37
  221. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/series/test_parity_sort.py +0 -38
  222. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/series/test_parity_stat.py +0 -38
  223. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_categorical.py +0 -66
  224. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_config.py +0 -37
  225. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_csv.py +0 -37
  226. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_dataframe_conversion.py +0 -42
  227. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_dataframe_spark_io.py +0 -39
  228. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_default_index.py +0 -49
  229. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_ewm.py +0 -37
  230. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_expanding.py +0 -39
  231. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_extension.py +0 -49
  232. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_frame_spark.py +0 -53
  233. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_generic_functions.py +0 -43
  234. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_indexing.py +0 -49
  235. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_indexops_spark.py +0 -39
  236. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_internal.py +0 -41
  237. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_namespace.py +0 -39
  238. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_numpy_compat.py +0 -60
  239. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_ops_on_diff_frames.py +0 -48
  240. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_ops_on_diff_frames_groupby.py +0 -39
  241. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_ops_on_diff_frames_groupby_expanding.py +0 -44
  242. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_ops_on_diff_frames_groupby_rolling.py +0 -84
  243. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_repr.py +0 -37
  244. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_resample.py +0 -45
  245. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_reshape.py +0 -39
  246. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_rolling.py +0 -39
  247. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_scalars.py +0 -37
  248. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_series_conversion.py +0 -39
  249. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_series_datetime.py +0 -39
  250. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_series_string.py +0 -39
  251. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_spark_functions.py +0 -39
  252. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_sql.py +0 -43
  253. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_stats.py +0 -37
  254. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_typedef.py +0 -36
  255. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_utils.py +0 -37
  256. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/connect/test_parity_window.py +0 -39
  257. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/data_type_ops/__init__.py +0 -16
  258. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/data_type_ops/test_base.py +0 -107
  259. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/data_type_ops/test_binary_ops.py +0 -224
  260. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/data_type_ops/test_boolean_ops.py +0 -825
  261. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/data_type_ops/test_categorical_ops.py +0 -562
  262. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/data_type_ops/test_complex_ops.py +0 -368
  263. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/data_type_ops/test_date_ops.py +0 -257
  264. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/data_type_ops/test_datetime_ops.py +0 -260
  265. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/data_type_ops/test_null_ops.py +0 -178
  266. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/data_type_ops/test_num_arithmetic.py +0 -184
  267. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/data_type_ops/test_num_ops.py +0 -497
  268. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/data_type_ops/test_num_reverse.py +0 -140
  269. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/data_type_ops/test_string_ops.py +0 -354
  270. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/data_type_ops/test_timedelta_ops.py +0 -219
  271. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/data_type_ops/test_udt_ops.py +0 -192
  272. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/data_type_ops/testing_utils.py +0 -228
  273. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/diff_frames_ops/__init__.py +0 -16
  274. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/diff_frames_ops/test_align.py +0 -118
  275. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/diff_frames_ops/test_basic_slow.py +0 -198
  276. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/diff_frames_ops/test_cov_corrwith.py +0 -181
  277. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/diff_frames_ops/test_dot_frame.py +0 -103
  278. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/diff_frames_ops/test_dot_series.py +0 -141
  279. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/diff_frames_ops/test_index.py +0 -109
  280. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/diff_frames_ops/test_series.py +0 -136
  281. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/diff_frames_ops/test_setitem_frame.py +0 -125
  282. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/diff_frames_ops/test_setitem_series.py +0 -217
  283. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/frame/__init__.py +0 -16
  284. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/frame/test_attrs.py +0 -384
  285. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/frame/test_constructor.py +0 -598
  286. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/frame/test_conversion.py +0 -73
  287. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/frame/test_reindexing.py +0 -869
  288. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/frame/test_reshaping.py +0 -487
  289. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/frame/test_spark.py +0 -309
  290. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/frame/test_take.py +0 -156
  291. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/frame/test_time_series.py +0 -149
  292. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/frame/test_truncate.py +0 -163
  293. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/groupby/__init__.py +0 -16
  294. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/groupby/test_aggregate.py +0 -311
  295. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/groupby/test_apply_func.py +0 -524
  296. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/groupby/test_cumulative.py +0 -419
  297. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/groupby/test_describe.py +0 -144
  298. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/groupby/test_groupby.py +0 -979
  299. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/groupby/test_head_tail.py +0 -234
  300. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/groupby/test_index.py +0 -206
  301. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/groupby/test_missing_data.py +0 -421
  302. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/groupby/test_split_apply.py +0 -187
  303. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/groupby/test_stat.py +0 -397
  304. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/indexes/__init__.py +0 -16
  305. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/indexes/test_align.py +0 -100
  306. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/indexes/test_base.py +0 -2743
  307. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/indexes/test_category.py +0 -484
  308. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/indexes/test_datetime.py +0 -276
  309. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/indexes/test_indexing.py +0 -432
  310. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/indexes/test_reindex.py +0 -310
  311. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/indexes/test_rename.py +0 -257
  312. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/indexes/test_reset_index.py +0 -160
  313. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/indexes/test_timedelta.py +0 -128
  314. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/io/__init__.py +0 -16
  315. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/io/test_io.py +0 -137
  316. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/plot/__init__.py +0 -16
  317. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/plot/test_frame_plot.py +0 -170
  318. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/plot/test_frame_plot_matplotlib.py +0 -547
  319. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/plot/test_frame_plot_plotly.py +0 -285
  320. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/plot/test_series_plot.py +0 -106
  321. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/plot/test_series_plot_matplotlib.py +0 -409
  322. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/plot/test_series_plot_plotly.py +0 -247
  323. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/series/__init__.py +0 -16
  324. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/series/test_all_any.py +0 -105
  325. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/series/test_arg_ops.py +0 -197
  326. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/series/test_as_of.py +0 -137
  327. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/series/test_as_type.py +0 -227
  328. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/series/test_compute.py +0 -634
  329. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/series/test_conversion.py +0 -88
  330. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/series/test_cumulative.py +0 -139
  331. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/series/test_index.py +0 -475
  332. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/series/test_missing_data.py +0 -265
  333. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/series/test_series.py +0 -818
  334. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/series/test_sort.py +0 -162
  335. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/series/test_stat.py +0 -780
  336. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_categorical.py +0 -741
  337. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_config.py +0 -160
  338. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_csv.py +0 -453
  339. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_dataframe_conversion.py +0 -281
  340. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_dataframe_spark_io.py +0 -487
  341. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_default_index.py +0 -109
  342. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_ewm.py +0 -434
  343. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_expanding.py +0 -253
  344. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_extension.py +0 -152
  345. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_frame_spark.py +0 -162
  346. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_generic_functions.py +0 -234
  347. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_indexing.py +0 -1339
  348. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_indexops_spark.py +0 -82
  349. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_internal.py +0 -124
  350. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_namespace.py +0 -638
  351. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_numpy_compat.py +0 -200
  352. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_ops_on_diff_frames.py +0 -1355
  353. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_ops_on_diff_frames_groupby.py +0 -655
  354. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_ops_on_diff_frames_groupby_expanding.py +0 -113
  355. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_ops_on_diff_frames_groupby_rolling.py +0 -118
  356. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_repr.py +0 -192
  357. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_resample.py +0 -346
  358. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_reshape.py +0 -495
  359. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_rolling.py +0 -263
  360. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_scalars.py +0 -59
  361. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_series_conversion.py +0 -85
  362. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_series_datetime.py +0 -364
  363. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_series_string.py +0 -362
  364. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_spark_functions.py +0 -46
  365. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_sql.py +0 -123
  366. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_stats.py +0 -581
  367. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_typedef.py +0 -447
  368. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_utils.py +0 -301
  369. snowflake/snowpark_connect/includes/python/pyspark/pandas/tests/test_window.py +0 -465
  370. snowflake/snowpark_connect/includes/python/pyspark/resource/tests/__init__.py +0 -16
  371. snowflake/snowpark_connect/includes/python/pyspark/resource/tests/test_resources.py +0 -83
  372. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/__init__.py +0 -16
  373. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/__init__.py +0 -16
  374. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/client/__init__.py +0 -16
  375. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/client/test_artifact.py +0 -420
  376. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/client/test_client.py +0 -358
  377. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/streaming/__init__.py +0 -16
  378. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/streaming/test_parity_foreach.py +0 -36
  379. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/streaming/test_parity_foreach_batch.py +0 -44
  380. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/streaming/test_parity_listener.py +0 -116
  381. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/streaming/test_parity_streaming.py +0 -35
  382. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_connect_basic.py +0 -3612
  383. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_connect_column.py +0 -1042
  384. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_connect_function.py +0 -2381
  385. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_connect_plan.py +0 -1060
  386. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_parity_arrow.py +0 -163
  387. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_parity_arrow_map.py +0 -38
  388. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_parity_arrow_python_udf.py +0 -48
  389. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_parity_catalog.py +0 -36
  390. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_parity_column.py +0 -55
  391. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_parity_conf.py +0 -36
  392. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_parity_dataframe.py +0 -96
  393. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_parity_datasources.py +0 -44
  394. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_parity_errors.py +0 -36
  395. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_parity_functions.py +0 -59
  396. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_parity_group.py +0 -36
  397. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_parity_pandas_cogrouped_map.py +0 -59
  398. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_parity_pandas_grouped_map.py +0 -74
  399. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_parity_pandas_grouped_map_with_state.py +0 -62
  400. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_parity_pandas_map.py +0 -58
  401. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_parity_pandas_udf.py +0 -70
  402. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_parity_pandas_udf_grouped_agg.py +0 -50
  403. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_parity_pandas_udf_scalar.py +0 -68
  404. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_parity_pandas_udf_window.py +0 -40
  405. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_parity_readwriter.py +0 -46
  406. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_parity_serde.py +0 -44
  407. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_parity_types.py +0 -100
  408. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_parity_udf.py +0 -100
  409. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_parity_udtf.py +0 -163
  410. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_session.py +0 -181
  411. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/connect/test_utils.py +0 -42
  412. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/pandas/__init__.py +0 -16
  413. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/pandas/test_pandas_cogrouped_map.py +0 -623
  414. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/pandas/test_pandas_grouped_map.py +0 -869
  415. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/pandas/test_pandas_grouped_map_with_state.py +0 -342
  416. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/pandas/test_pandas_map.py +0 -436
  417. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/pandas/test_pandas_udf.py +0 -363
  418. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/pandas/test_pandas_udf_grouped_agg.py +0 -592
  419. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/pandas/test_pandas_udf_scalar.py +0 -1503
  420. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/pandas/test_pandas_udf_typehints.py +0 -392
  421. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/pandas/test_pandas_udf_typehints_with_future_annotations.py +0 -375
  422. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/pandas/test_pandas_udf_window.py +0 -411
  423. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/streaming/__init__.py +0 -16
  424. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/streaming/test_streaming.py +0 -401
  425. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/streaming/test_streaming_foreach.py +0 -295
  426. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/streaming/test_streaming_foreach_batch.py +0 -106
  427. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/streaming/test_streaming_listener.py +0 -558
  428. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/test_arrow.py +0 -1346
  429. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/test_arrow_map.py +0 -182
  430. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/test_arrow_python_udf.py +0 -202
  431. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/test_catalog.py +0 -503
  432. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/test_column.py +0 -225
  433. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/test_conf.py +0 -83
  434. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/test_context.py +0 -201
  435. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/test_dataframe.py +0 -1931
  436. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/test_datasources.py +0 -256
  437. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/test_errors.py +0 -69
  438. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/test_functions.py +0 -1349
  439. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/test_group.py +0 -53
  440. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/test_pandas_sqlmetrics.py +0 -68
  441. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/test_readwriter.py +0 -283
  442. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/test_serde.py +0 -155
  443. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/test_session.py +0 -412
  444. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/test_types.py +0 -1581
  445. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/test_udf.py +0 -961
  446. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/test_udf_profiler.py +0 -165
  447. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/test_udtf.py +0 -1456
  448. snowflake/snowpark_connect/includes/python/pyspark/sql/tests/test_utils.py +0 -1686
  449. snowflake/snowpark_connect/includes/python/pyspark/streaming/tests/__init__.py +0 -16
  450. snowflake/snowpark_connect/includes/python/pyspark/streaming/tests/test_context.py +0 -184
  451. snowflake/snowpark_connect/includes/python/pyspark/streaming/tests/test_dstream.py +0 -706
  452. snowflake/snowpark_connect/includes/python/pyspark/streaming/tests/test_kinesis.py +0 -118
  453. snowflake/snowpark_connect/includes/python/pyspark/streaming/tests/test_listener.py +0 -160
  454. snowflake/snowpark_connect/includes/python/pyspark/tests/__init__.py +0 -16
  455. snowflake/snowpark_connect/includes/python/pyspark/tests/test_appsubmit.py +0 -306
  456. snowflake/snowpark_connect/includes/python/pyspark/tests/test_broadcast.py +0 -196
  457. snowflake/snowpark_connect/includes/python/pyspark/tests/test_conf.py +0 -44
  458. snowflake/snowpark_connect/includes/python/pyspark/tests/test_context.py +0 -346
  459. snowflake/snowpark_connect/includes/python/pyspark/tests/test_daemon.py +0 -89
  460. snowflake/snowpark_connect/includes/python/pyspark/tests/test_install_spark.py +0 -124
  461. snowflake/snowpark_connect/includes/python/pyspark/tests/test_join.py +0 -69
  462. snowflake/snowpark_connect/includes/python/pyspark/tests/test_memory_profiler.py +0 -167
  463. snowflake/snowpark_connect/includes/python/pyspark/tests/test_pin_thread.py +0 -194
  464. snowflake/snowpark_connect/includes/python/pyspark/tests/test_profiler.py +0 -168
  465. snowflake/snowpark_connect/includes/python/pyspark/tests/test_rdd.py +0 -939
  466. snowflake/snowpark_connect/includes/python/pyspark/tests/test_rddbarrier.py +0 -52
  467. snowflake/snowpark_connect/includes/python/pyspark/tests/test_rddsampler.py +0 -66
  468. snowflake/snowpark_connect/includes/python/pyspark/tests/test_readwrite.py +0 -368
  469. snowflake/snowpark_connect/includes/python/pyspark/tests/test_serializers.py +0 -257
  470. snowflake/snowpark_connect/includes/python/pyspark/tests/test_shuffle.py +0 -267
  471. snowflake/snowpark_connect/includes/python/pyspark/tests/test_stage_sched.py +0 -153
  472. snowflake/snowpark_connect/includes/python/pyspark/tests/test_statcounter.py +0 -130
  473. snowflake/snowpark_connect/includes/python/pyspark/tests/test_taskcontext.py +0 -350
  474. snowflake/snowpark_connect/includes/python/pyspark/tests/test_util.py +0 -97
  475. snowflake/snowpark_connect/includes/python/pyspark/tests/test_worker.py +0 -271
  476. snowpark_connect-0.24.0.dist-info/RECORD +0 -898
  477. {snowpark_connect-0.24.0.data → snowpark_connect-0.26.0.data}/scripts/snowpark-connect +0 -0
  478. {snowpark_connect-0.24.0.data → snowpark_connect-0.26.0.data}/scripts/snowpark-session +0 -0
  479. {snowpark_connect-0.24.0.data → snowpark_connect-0.26.0.data}/scripts/snowpark-submit +0 -0
  480. {snowpark_connect-0.24.0.dist-info → snowpark_connect-0.26.0.dist-info}/WHEEL +0 -0
  481. {snowpark_connect-0.24.0.dist-info → snowpark_connect-0.26.0.dist-info}/licenses/LICENSE-binary +0 -0
  482. {snowpark_connect-0.24.0.dist-info → snowpark_connect-0.26.0.dist-info}/licenses/LICENSE.txt +0 -0
  483. {snowpark_connect-0.24.0.dist-info → snowpark_connect-0.26.0.dist-info}/licenses/NOTICE-binary +0 -0
  484. {snowpark_connect-0.24.0.dist-info → snowpark_connect-0.26.0.dist-info}/top_level.txt +0 -0
@@ -1,401 +0,0 @@
1
- #
2
- # Licensed to the Apache Software Foundation (ASF) under one or more
3
- # contributor license agreements. See the NOTICE file distributed with
4
- # this work for additional information regarding copyright ownership.
5
- # The ASF licenses this file to You under the Apache License, Version 2.0
6
- # (the "License"); you may not use this file except in compliance with
7
- # the License. You may obtain a copy of the License at
8
- #
9
- # http://www.apache.org/licenses/LICENSE-2.0
10
- #
11
- # Unless required by applicable law or agreed to in writing, software
12
- # distributed under the License is distributed on an "AS IS" BASIS,
13
- # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
14
- # See the License for the specific language governing permissions and
15
- # limitations under the License.
16
- #
17
-
18
- import os
19
- import shutil
20
- import tempfile
21
- import time
22
-
23
- from pyspark.sql import Row
24
- from pyspark.sql.functions import lit
25
- from pyspark.sql.types import StructType, StructField, IntegerType, StringType
26
- from pyspark.testing.sqlutils import ReusedSQLTestCase
27
- from pyspark.errors.exceptions.connect import SparkConnectException
28
-
29
-
30
- class StreamingTestsMixin:
31
- def test_streaming_query_functions_basic(self):
32
- df = self.spark.readStream.format("rate").option("rowsPerSecond", 10).load()
33
- query = (
34
- df.writeStream.format("memory")
35
- .queryName("test_streaming_query_functions_basic")
36
- .start()
37
- )
38
- try:
39
- self.assertEquals(query.name, "test_streaming_query_functions_basic")
40
- self.assertTrue(isinstance(query.id, str))
41
- self.assertTrue(isinstance(query.runId, str))
42
- self.assertTrue(query.isActive)
43
- self.assertEqual(query.exception(), None)
44
- self.assertFalse(query.awaitTermination(1))
45
- query.processAllAvailable()
46
- recentProgress = query.recentProgress
47
- lastProgress = query.lastProgress
48
- self.assertEqual(lastProgress["name"], query.name)
49
- self.assertEqual(lastProgress["id"], query.id)
50
- self.assertTrue(any(p == lastProgress for p in recentProgress))
51
- query.explain()
52
-
53
- except Exception as e:
54
- self.fail(
55
- "Streaming query functions sanity check shouldn't throw any error. "
56
- "Error message: " + str(e)
57
- )
58
-
59
- finally:
60
- query.stop()
61
-
62
- def test_stream_trigger(self):
63
- df = self.spark.readStream.format("text").load("python/test_support/sql/streaming")
64
-
65
- # Should take at least one arg
66
- try:
67
- df.writeStream.trigger()
68
- except ValueError:
69
- pass
70
-
71
- # Should not take multiple args
72
- try:
73
- df.writeStream.trigger(once=True, processingTime="5 seconds")
74
- except ValueError:
75
- pass
76
-
77
- # Should not take multiple args
78
- try:
79
- df.writeStream.trigger(processingTime="5 seconds", continuous="1 second")
80
- except ValueError:
81
- pass
82
-
83
- # Should take only keyword args
84
- try:
85
- df.writeStream.trigger("5 seconds")
86
- self.fail("Should have thrown an exception")
87
- except TypeError:
88
- pass
89
-
90
- def test_stream_read_options(self):
91
- schema = StructType([StructField("data", StringType(), False)])
92
- df = (
93
- self.spark.readStream.format("text")
94
- .option("path", "python/test_support/sql/streaming")
95
- .schema(schema)
96
- .load()
97
- )
98
- self.assertTrue(df.isStreaming)
99
- self.assertEqual(df.schema.simpleString(), "struct<data:string>")
100
-
101
- def test_stream_read_options_overwrite(self):
102
- bad_schema = StructType([StructField("test", IntegerType(), False)])
103
- schema = StructType([StructField("data", StringType(), False)])
104
- # SPARK-32516 disables the overwrite behavior by default.
105
- with self.sql_conf({"spark.sql.legacy.pathOptionBehavior.enabled": True}):
106
- df = (
107
- self.spark.readStream.format("csv")
108
- .option("path", "python/test_support/sql/fake")
109
- .schema(bad_schema)
110
- .load(path="python/test_support/sql/streaming", schema=schema, format="text")
111
- )
112
- self.assertTrue(df.isStreaming)
113
- self.assertEqual(df.schema.simpleString(), "struct<data:string>")
114
-
115
- def test_stream_save_options(self):
116
- df = (
117
- self.spark.readStream.format("text")
118
- .load("python/test_support/sql/streaming")
119
- .withColumn("id", lit(1))
120
- )
121
- for q in self.spark.streams.active:
122
- q.stop()
123
- tmpPath = tempfile.mkdtemp()
124
- shutil.rmtree(tmpPath)
125
- self.assertTrue(df.isStreaming)
126
- out = os.path.join(tmpPath, "out")
127
- chk = os.path.join(tmpPath, "chk")
128
- q = (
129
- df.writeStream.option("checkpointLocation", chk)
130
- .queryName("this_query")
131
- .format("parquet")
132
- .partitionBy("id")
133
- .outputMode("append")
134
- .option("path", out)
135
- .start()
136
- )
137
- try:
138
- self.assertEqual(q.name, "this_query")
139
- self.assertTrue(q.isActive)
140
- q.processAllAvailable()
141
- output_files = []
142
- for _, _, files in os.walk(out):
143
- output_files.extend([f for f in files if not f.startswith(".")])
144
- self.assertTrue(len(output_files) > 0)
145
- self.assertTrue(len(os.listdir(chk)) > 0)
146
- finally:
147
- q.stop()
148
- shutil.rmtree(tmpPath)
149
-
150
- def test_stream_save_options_overwrite(self):
151
- df = self.spark.readStream.format("text").load("python/test_support/sql/streaming")
152
- for q in self.spark.streams.active:
153
- q.stop()
154
- tmpPath = tempfile.mkdtemp()
155
- shutil.rmtree(tmpPath)
156
- self.assertTrue(df.isStreaming)
157
- out = os.path.join(tmpPath, "out")
158
- chk = os.path.join(tmpPath, "chk")
159
- fake1 = os.path.join(tmpPath, "fake1")
160
- fake2 = os.path.join(tmpPath, "fake2")
161
- # SPARK-32516 disables the overwrite behavior by default.
162
- with self.sql_conf({"spark.sql.legacy.pathOptionBehavior.enabled": True}):
163
- q = (
164
- df.writeStream.option("checkpointLocation", fake1)
165
- .format("memory")
166
- .option("path", fake2)
167
- .queryName("fake_query")
168
- .outputMode("append")
169
- .start(path=out, format="parquet", queryName="this_query", checkpointLocation=chk)
170
- )
171
-
172
- try:
173
- self.assertEqual(q.name, "this_query")
174
- self.assertTrue(q.isActive)
175
- q.processAllAvailable()
176
- output_files = []
177
- for _, _, files in os.walk(out):
178
- output_files.extend([f for f in files if not f.startswith(".")])
179
- self.assertTrue(len(output_files) > 0)
180
- self.assertTrue(len(os.listdir(chk)) > 0)
181
- self.assertFalse(os.path.isdir(fake1)) # should not have been created
182
- self.assertFalse(os.path.isdir(fake2)) # should not have been created
183
- finally:
184
- q.stop()
185
- shutil.rmtree(tmpPath)
186
-
187
- def test_stream_status_and_progress(self):
188
- df = self.spark.readStream.format("text").load("python/test_support/sql/streaming")
189
- for q in self.spark.streams.active:
190
- q.stop()
191
- tmpPath = tempfile.mkdtemp()
192
- shutil.rmtree(tmpPath)
193
- self.assertTrue(df.isStreaming)
194
- out = os.path.join(tmpPath, "out")
195
- chk = os.path.join(tmpPath, "chk")
196
-
197
- def func(x):
198
- time.sleep(1)
199
- return x
200
-
201
- from pyspark.sql.functions import col, udf
202
-
203
- sleep_udf = udf(func)
204
-
205
- # Use "sleep_udf" to delay the progress update so that we can test `lastProgress` when there
206
- # were no updates.
207
- q = df.select(sleep_udf(col("value")).alias("value")).writeStream.start(
208
- path=out, format="parquet", queryName="this_query", checkpointLocation=chk
209
- )
210
- try:
211
- # "lastProgress" will return None in most cases. However, as it may be flaky when
212
- # Jenkins is very slow, we don't assert it. If there is something wrong, "lastProgress"
213
- # may throw error with a high chance and make this test flaky, so we should still be
214
- # able to detect broken codes.
215
- q.lastProgress
216
-
217
- q.processAllAvailable()
218
- lastProgress = q.lastProgress
219
- recentProgress = q.recentProgress
220
- status = q.status
221
- self.assertEqual(lastProgress["name"], q.name)
222
- self.assertEqual(lastProgress["id"], q.id)
223
- self.assertTrue(any(p == lastProgress for p in recentProgress))
224
- self.assertTrue(
225
- "message" in status and "isDataAvailable" in status and "isTriggerActive" in status
226
- )
227
- finally:
228
- q.stop()
229
- shutil.rmtree(tmpPath)
230
-
231
- def test_stream_await_termination(self):
232
- df = self.spark.readStream.format("text").load("python/test_support/sql/streaming")
233
- for q in self.spark.streams.active:
234
- q.stop()
235
- tmpPath = tempfile.mkdtemp()
236
- shutil.rmtree(tmpPath)
237
- self.assertTrue(df.isStreaming)
238
- out = os.path.join(tmpPath, "out")
239
- chk = os.path.join(tmpPath, "chk")
240
- q = df.writeStream.start(
241
- path=out, format="parquet", queryName="this_query", checkpointLocation=chk
242
- )
243
- try:
244
- self.assertTrue(q.isActive)
245
- try:
246
- q.awaitTermination("hello")
247
- self.fail("Expected a value exception")
248
- except ValueError:
249
- pass
250
- now = time.time()
251
- # test should take at least 2 seconds
252
- res = q.awaitTermination(2.6)
253
- duration = time.time() - now
254
- self.assertTrue(duration >= 2)
255
- self.assertFalse(res)
256
-
257
- q.processAllAvailable()
258
- q.stop()
259
- # Sanity check when no parameter is set
260
- q.awaitTermination()
261
- self.assertFalse(q.isActive)
262
- finally:
263
- q.stop()
264
- shutil.rmtree(tmpPath)
265
-
266
- def test_stream_exception(self):
267
- with self.sql_conf({"spark.sql.execution.pyspark.udf.simplifiedTraceback.enabled": True}):
268
- sdf = self.spark.readStream.format("text").load("python/test_support/sql/streaming")
269
- sq = sdf.writeStream.format("memory").queryName("query_explain").start()
270
- try:
271
- sq.processAllAvailable()
272
- self.assertEqual(sq.exception(), None)
273
- finally:
274
- sq.stop()
275
-
276
- from pyspark.sql.functions import col, udf
277
- from pyspark.errors import StreamingQueryException
278
-
279
- bad_udf = udf(lambda x: 1 / 0)
280
- sq = (
281
- sdf.select(bad_udf(col("value")))
282
- .writeStream.format("memory")
283
- .queryName("this_query")
284
- .start()
285
- )
286
- try:
287
- # Process some data to fail the query
288
- sq.processAllAvailable()
289
- self.fail("bad udf should fail the query")
290
- except StreamingQueryException as e:
291
- # This is expected
292
- self._assert_exception_tree_contains_msg(e, "ZeroDivisionError")
293
- finally:
294
- exception = sq.exception()
295
- sq.stop()
296
- self.assertIsInstance(exception, StreamingQueryException)
297
- self._assert_exception_tree_contains_msg(exception, "ZeroDivisionError")
298
-
299
- def _assert_exception_tree_contains_msg(self, exception, msg):
300
- if isinstance(exception, SparkConnectException):
301
- self._assert_exception_tree_contains_msg_connect(exception, msg)
302
- else:
303
- self._assert_exception_tree_contains_msg_default(exception, msg)
304
-
305
- def _assert_exception_tree_contains_msg_connect(self, exception, msg):
306
- self.assertTrue(
307
- msg in exception.message,
308
- "Exception tree doesn't contain the expected message: %s" % msg,
309
- )
310
-
311
- def _assert_exception_tree_contains_msg_default(self, exception, msg):
312
- e = exception
313
- contains = msg in e.desc
314
- while e.cause is not None and not contains:
315
- e = e.cause
316
- contains = msg in e.desc
317
- self.assertTrue(contains, "Exception tree doesn't contain the expected message: %s" % msg)
318
-
319
- def test_query_manager_get(self):
320
- df = self.spark.readStream.format("rate").load()
321
- for q in self.spark.streams.active:
322
- q.stop()
323
- q = df.writeStream.format("noop").start()
324
-
325
- self.assertTrue(q.isActive)
326
- self.assertTrue(q.id == self.spark.streams.get(q.id).id)
327
-
328
- q.stop()
329
-
330
- self.assertIsNone(self.spark.streams.get(q.id))
331
-
332
- def test_query_manager_await_termination(self):
333
- df = self.spark.readStream.format("text").load("python/test_support/sql/streaming")
334
- for q in self.spark.streams.active:
335
- q.stop()
336
- tmpPath = tempfile.mkdtemp()
337
- shutil.rmtree(tmpPath)
338
- self.assertTrue(df.isStreaming)
339
- out = os.path.join(tmpPath, "out")
340
- chk = os.path.join(tmpPath, "chk")
341
- q = df.writeStream.start(
342
- path=out, format="parquet", queryName="this_query", checkpointLocation=chk
343
- )
344
- try:
345
- self.assertTrue(q.isActive)
346
- try:
347
- self.spark.streams.awaitAnyTermination("hello")
348
- self.fail("Expected a value exception")
349
- except ValueError:
350
- pass
351
- now = time.time()
352
- # test should take at least 2 seconds
353
- res = self.spark.streams.awaitAnyTermination(2.6)
354
- duration = time.time() - now
355
- self.assertTrue(duration >= 2)
356
- self.assertFalse(res)
357
- finally:
358
- q.processAllAvailable()
359
- q.stop()
360
- shutil.rmtree(tmpPath)
361
-
362
- def test_streaming_read_from_table(self):
363
- with self.table("input_table", "this_query"):
364
- self.spark.sql("CREATE TABLE input_table (value string) USING parquet")
365
- self.spark.sql("INSERT INTO input_table VALUES ('aaa'), ('bbb'), ('ccc')")
366
- df = self.spark.readStream.table("input_table")
367
- self.assertTrue(df.isStreaming)
368
- q = df.writeStream.format("memory").queryName("this_query").start()
369
- q.processAllAvailable()
370
- q.stop()
371
- result = self.spark.sql("SELECT * FROM this_query ORDER BY value").collect()
372
- self.assertEqual(
373
- set([Row(value="aaa"), Row(value="bbb"), Row(value="ccc")]), set(result)
374
- )
375
-
376
- def test_streaming_write_to_table(self):
377
- with self.table("output_table"), tempfile.TemporaryDirectory() as tmpdir:
378
- df = self.spark.readStream.format("rate").option("rowsPerSecond", 10).load()
379
- q = df.writeStream.toTable("output_table", format="parquet", checkpointLocation=tmpdir)
380
- self.assertTrue(q.isActive)
381
- time.sleep(10)
382
- q.stop()
383
- result = self.spark.sql("SELECT value FROM output_table").collect()
384
- self.assertTrue(len(result) > 0)
385
-
386
-
387
- class StreamingTests(StreamingTestsMixin, ReusedSQLTestCase):
388
- pass
389
-
390
-
391
- if __name__ == "__main__":
392
- import unittest
393
- from pyspark.sql.tests.streaming.test_streaming import * # noqa: F401
394
-
395
- try:
396
- import xmlrunner
397
-
398
- testRunner = xmlrunner.XMLTestRunner(output="target/test-reports", verbosity=2)
399
- except ImportError:
400
- testRunner = None
401
- unittest.main(testRunner=testRunner, verbosity=2)
@@ -1,295 +0,0 @@
1
- #
2
- # Licensed to the Apache Software Foundation (ASF) under one or more
3
- # contributor license agreements. See the NOTICE file distributed with
4
- # this work for additional information regarding copyright ownership.
5
- # The ASF licenses this file to You under the Apache License, Version 2.0
6
- # (the "License"); you may not use this file except in compliance with
7
- # the License. You may obtain a copy of the License at
8
- #
9
- # http://www.apache.org/licenses/LICENSE-2.0
10
- #
11
- # Unless required by applicable law or agreed to in writing, software
12
- # distributed under the License is distributed on an "AS IS" BASIS,
13
- # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
14
- # See the License for the specific language governing permissions and
15
- # limitations under the License.
16
- #
17
-
18
- import os
19
- import tempfile
20
-
21
- from pyspark.testing.sqlutils import ReusedSQLTestCase
22
-
23
-
24
- class StreamingTestsForeachMixin:
25
- class ForeachWriterTester:
26
- def __init__(self, spark):
27
- self.spark = spark
28
-
29
- def write_open_event(self, partitionId, epochId):
30
- self._write_event(self.open_events_dir, {"partition": partitionId, "epoch": epochId})
31
-
32
- def write_process_event(self, row):
33
- self._write_event(self.process_events_dir, {"value": "text"})
34
-
35
- def write_close_event(self, error):
36
- self._write_event(self.close_events_dir, {"error": str(error)})
37
-
38
- def write_input_file(self):
39
- self._write_event(self.input_dir, "text")
40
-
41
- def open_events(self):
42
- return self._read_events(self.open_events_dir, "partition INT, epoch INT")
43
-
44
- def process_events(self):
45
- return self._read_events(self.process_events_dir, "value STRING")
46
-
47
- def close_events(self):
48
- return self._read_events(self.close_events_dir, "error STRING")
49
-
50
- def run_streaming_query_on_writer(self, writer, num_files):
51
- self._reset()
52
- try:
53
- sdf = self.spark.readStream.format("text").load(self.input_dir)
54
- sq = sdf.writeStream.foreach(writer).start()
55
- for i in range(num_files):
56
- self.write_input_file()
57
- sq.processAllAvailable()
58
- finally:
59
- self.stop_all()
60
-
61
- def assert_invalid_writer(self, writer, msg=None):
62
- self._reset()
63
- try:
64
- sdf = self.spark.readStream.format("text").load(self.input_dir)
65
- sq = sdf.writeStream.foreach(writer).start()
66
- self.write_input_file()
67
- sq.processAllAvailable()
68
- self.fail("invalid writer %s did not fail the query" % str(writer)) # not expected
69
- except Exception as e:
70
- if msg:
71
- assert msg in str(e), "%s not in %s" % (msg, str(e))
72
-
73
- finally:
74
- self.stop_all()
75
-
76
- def stop_all(self):
77
- for q in self.spark.streams.active:
78
- q.stop()
79
-
80
- def _reset(self):
81
- self.input_dir = tempfile.mkdtemp()
82
- self.open_events_dir = tempfile.mkdtemp()
83
- self.process_events_dir = tempfile.mkdtemp()
84
- self.close_events_dir = tempfile.mkdtemp()
85
-
86
- def _read_events(self, dir, json):
87
- rows = self.spark.read.schema(json).json(dir).collect()
88
- dicts = [row.asDict() for row in rows]
89
- return dicts
90
-
91
- def _write_event(self, dir, event):
92
- import uuid
93
-
94
- with open(os.path.join(dir, str(uuid.uuid4())), "w") as f:
95
- f.write("%s\n" % str(event))
96
-
97
- def __getstate__(self):
98
- return (self.open_events_dir, self.process_events_dir, self.close_events_dir)
99
-
100
- def __setstate__(self, state):
101
- self.open_events_dir, self.process_events_dir, self.close_events_dir = state
102
-
103
- # Those foreach tests are failed in macOS High Sierra by defined rules
104
- # at http://sealiesoftware.com/blog/archive/2017/6/5/Objective-C_and_fork_in_macOS_1013.html
105
- # To work around this, OBJC_DISABLE_INITIALIZE_FORK_SAFETY=YES.
106
- def test_streaming_foreach_with_simple_function(self):
107
- tester = self.ForeachWriterTester(self.spark)
108
-
109
- def foreach_func(row):
110
- tester.write_process_event(row)
111
-
112
- tester.run_streaming_query_on_writer(foreach_func, 2)
113
- self.assertEqual(len(tester.process_events()), 2)
114
-
115
- def test_streaming_foreach_with_basic_open_process_close(self):
116
- tester = self.ForeachWriterTester(self.spark)
117
-
118
- class ForeachWriter:
119
- def open(self, partitionId, epochId):
120
- tester.write_open_event(partitionId, epochId)
121
- return True
122
-
123
- def process(self, row):
124
- tester.write_process_event(row)
125
-
126
- def close(self, error):
127
- tester.write_close_event(error)
128
-
129
- tester.run_streaming_query_on_writer(ForeachWriter(), 2)
130
-
131
- open_events = tester.open_events()
132
- self.assertEqual(len(open_events), 2)
133
- self.assertSetEqual(set([e["epoch"] for e in open_events]), {0, 1})
134
-
135
- self.assertEqual(len(tester.process_events()), 2)
136
-
137
- close_events = tester.close_events()
138
- self.assertEqual(len(close_events), 2)
139
- self.assertSetEqual(set([e["error"] for e in close_events]), {"None"})
140
-
141
- def test_streaming_foreach_with_open_returning_false(self):
142
- tester = self.ForeachWriterTester(self.spark)
143
-
144
- class ForeachWriter:
145
- def open(self, partition_id, epoch_id):
146
- tester.write_open_event(partition_id, epoch_id)
147
- return False
148
-
149
- def process(self, row):
150
- tester.write_process_event(row)
151
-
152
- def close(self, error):
153
- tester.write_close_event(error)
154
-
155
- tester.run_streaming_query_on_writer(ForeachWriter(), 2)
156
-
157
- self.assertEqual(len(tester.open_events()), 2)
158
-
159
- self.assertEqual(len(tester.process_events()), 0) # no row was processed
160
-
161
- close_events = tester.close_events()
162
- self.assertEqual(len(close_events), 2)
163
- self.assertSetEqual(set([e["error"] for e in close_events]), {"None"})
164
-
165
- def test_streaming_foreach_without_open_method(self):
166
- tester = self.ForeachWriterTester(self.spark)
167
-
168
- class ForeachWriter:
169
- def process(self, row):
170
- tester.write_process_event(row)
171
-
172
- def close(self, error):
173
- tester.write_close_event(error)
174
-
175
- tester.run_streaming_query_on_writer(ForeachWriter(), 2)
176
- self.assertEqual(len(tester.open_events()), 0) # no open events
177
- self.assertEqual(len(tester.process_events()), 2)
178
- self.assertEqual(len(tester.close_events()), 2)
179
-
180
- def test_streaming_foreach_without_close_method(self):
181
- tester = self.ForeachWriterTester(self.spark)
182
-
183
- class ForeachWriter:
184
- def open(self, partition_id, epoch_id):
185
- tester.write_open_event(partition_id, epoch_id)
186
- return True
187
-
188
- def process(self, row):
189
- tester.write_process_event(row)
190
-
191
- tester.run_streaming_query_on_writer(ForeachWriter(), 2)
192
- self.assertEqual(len(tester.open_events()), 2) # no open events
193
- self.assertEqual(len(tester.process_events()), 2)
194
- self.assertEqual(len(tester.close_events()), 0)
195
-
196
- def test_streaming_foreach_without_open_and_close_methods(self):
197
- tester = self.ForeachWriterTester(self.spark)
198
-
199
- class ForeachWriter:
200
- def process(self, row):
201
- tester.write_process_event(row)
202
-
203
- tester.run_streaming_query_on_writer(ForeachWriter(), 2)
204
- self.assertEqual(len(tester.open_events()), 0) # no open events
205
- self.assertEqual(len(tester.process_events()), 2)
206
- self.assertEqual(len(tester.close_events()), 0)
207
-
208
- def test_streaming_foreach_with_process_throwing_error(self):
209
- from pyspark.errors import StreamingQueryException
210
-
211
- tester = self.ForeachWriterTester(self.spark)
212
-
213
- class ForeachWriter:
214
- def process(self, row):
215
- raise RuntimeError("test error")
216
-
217
- def close(self, error):
218
- tester.write_close_event(error)
219
-
220
- try:
221
- tester.run_streaming_query_on_writer(ForeachWriter(), 1)
222
- self.fail("bad writer did not fail the query") # this is not expected
223
- except StreamingQueryException:
224
- # TODO: Verify whether original error message is inside the exception
225
- pass
226
-
227
- self.assertEqual(len(tester.process_events()), 0) # no row was processed
228
- close_events = tester.close_events()
229
- self.assertEqual(len(close_events), 1)
230
- # TODO: Verify whether original error message is inside the exception
231
-
232
- def test_streaming_foreach_with_invalid_writers(self):
233
-
234
- tester = self.ForeachWriterTester(self.spark)
235
-
236
- def func_with_iterator_input(iter):
237
- for x in iter:
238
- print(x)
239
-
240
- tester.assert_invalid_writer(func_with_iterator_input)
241
-
242
- class WriterWithoutProcess:
243
- def open(self, partition):
244
- pass
245
-
246
- tester.assert_invalid_writer(WriterWithoutProcess(), "ATTRIBUTE_NOT_CALLABLE")
247
-
248
- class WriterWithNonCallableProcess:
249
- process = True
250
-
251
- tester.assert_invalid_writer(WriterWithNonCallableProcess(), "ATTRIBUTE_NOT_CALLABLE")
252
-
253
- class WriterWithNoParamProcess:
254
- def process(self):
255
- pass
256
-
257
- tester.assert_invalid_writer(WriterWithNoParamProcess())
258
-
259
- # Abstract class for tests below
260
- class WithProcess:
261
- def process(self, row):
262
- pass
263
-
264
- class WriterWithNonCallableOpen(WithProcess):
265
- open = True
266
-
267
- tester.assert_invalid_writer(WriterWithNonCallableOpen(), "ATTRIBUTE_NOT_CALLABLE")
268
-
269
- class WriterWithNoParamOpen(WithProcess):
270
- def open(self):
271
- pass
272
-
273
- tester.assert_invalid_writer(WriterWithNoParamOpen())
274
-
275
- class WriterWithNonCallableClose(WithProcess):
276
- close = True
277
-
278
- tester.assert_invalid_writer(WriterWithNonCallableClose(), "ATTRIBUTE_NOT_CALLABLE")
279
-
280
-
281
- class StreamingTestsForeach(StreamingTestsForeachMixin, ReusedSQLTestCase):
282
- pass
283
-
284
-
285
- if __name__ == "__main__":
286
- import unittest
287
- from pyspark.sql.tests.streaming.test_streaming_foreach import * # noqa: F401
288
-
289
- try:
290
- import xmlrunner
291
-
292
- testRunner = xmlrunner.XMLTestRunner(output="target/test-reports", verbosity=2)
293
- except ImportError:
294
- testRunner = None
295
- unittest.main(testRunner=testRunner, verbosity=2)